Comparing Predictive Accuracy
- 1 July 1995
- journal article
- research article
- Published by Taylor & Francis in Journal of Business & Economic Statistics
- Vol. 13 (3) , 253-263
- https://doi.org/10.1080/07350015.1995.10524599
Abstract
We propose and evaluate explicit tests of the null hypothesis of no difference in the accuracy of two competing forecasts. In contrast to previously developed tests, a wide variety of accuracy measures can be used (in particular, the loss function need not be quadratic and need not even be symmetric), and forecast errors can be non-Gaussian, nonzero mean, serially correlated, and contemporaneously correlated. Asymptotic and exact finite-sample tests are proposed, evaluated, and illustrated.Keywords
All Related Versions
This publication has 26 references indexed in Scilit:
- Federal Budget Projections: A Nonparametric Assessment of Bias and EfficiencyThe Review of Economics and Statistics, 1995
- Forecasting Output with the Composite Leading Index: A Real-Time AnalysisJournal of the American Statistical Association, 1991
- Multi-step estimation and forecasting in dynamic modelsJournal of Econometrics, 1991
- Posterior, predictive, and utility-based approaches to testing the arbitrage pricing theoryJournal of Financial Economics, 1990
- Unit roots in real GNP: Do we know, and do we care?Carnegie-Rochester Conference Series on Public Policy, 1990
- Likelihood Ratio Tests for Model Selection and Non-Nested HypothesesEconometrica, 1989
- Interpreting the evidence on money-income causalityJournal of Econometrics, 1989
- Estimating Time Series Models Using the Relevant Forecast Evaluation CriterionJournal of the Royal Statistical Society. Series A (General), 1984
- Notes on Testing the Predictive Performance of Econometric ModelsInternational Economic Review, 1974
- Prediction with a Generalized Cost of Error FunctionJournal of the Operational Research Society, 1969